Tips for Building a

Data Science Capability

Some Hard-Won Best Practices and (Surprising) Lessons Learned
2 | Introduction
CONTENTS

INTRODUCTION...................................................................................... V

Analytics Culture
W H Y H A S N T Y O U R D ATA S C I E N C E I N V E S T M E N T
D E L I V E R E D O N I T S P R O M I S E ? ...................................................... 6

Organizational Profiles
G E T T I N G PA S T T H E B U M P S I N T H E R O A D .................................. 13

Analytics and Change

K E Y S T O B U I L D I N G B U Y- I N ........................................................... 17

Aligning Data Science

M A K I N G O R G A N I Z AT I O N A L S T R U C T U R E W O R K ......................... 22

The Leadership Angle

H A R N E S S I N G T H E P O W E R O F D ATA T H R O U G H
T H E S TA N D - U P O F A C H I E F D ATA O F F I C E R ................................ 27

Everything you Need to Know About Managing Your Data Science Talent
T H E B O O Z A L L E N D ATA S C I E N C E TA L E N T
M A N A G E M E N T M O D E L .................................................................... 31

The Data Science Challenge

H OW D ES I G N T HI N K I N G CA N H EL P YOU R E A L IZE
O R G A N I Z AT I O N A L V A L U E ............................................................... 36

CONCLUSION........................................................................................ 42

Contents | iii
iv | Introduction
 INTRODUCTION

Many organizations believe in the power and potential of data science

but are challenged in establishing a sustainable data science capability.
How do organizations embed data science across their enterprise so
that it can deliver the next level of organizational performance and return
on investment?

Building a data science capability in any organization isnt easytheres

a lot to learn, with roadblocks and pitfalls at every turn. But it can be
doneand done right. This booklet will show you how. Weve filled it with
the most valuable best practices and lessons weve learnedboth in our
pioneering work with clients, and in building our own, 500-member data
science team, one of the worlds largest.

In these pages, youll discover:
++ W
 hy your data science investment may not
have delivered on its promise so far. It could
be that you havent developed an analytics-
driven culture.
++ How to keep from getting bogged down. data evangelist.
Some organizations have difficulty prioritizing
their data science projects. Others face skepti-
cism from leadership. Heres how to overcome
the most common roadblocks.
++ How to get buy-in for data science throughout
the entire organization. We show you how to
overcome resistance to everything from using
analytics to sharing information.
++ Where to place your data science teams in your
organization. Should they be centralized?
Dispersed? Permanently embedded in individual
business units? Each option has its advantages
and risks.
++ How to stand up the position of Chief Data
Officer (CDO). We tell you why a CDO needs
to be one part enforcer and two parts
++ How you can leverage our Data Science Talent
Management Model. Well help you answer three
key questions: Who do you need? Where do you
need them? How do you keep them?
++ Why design thinking-when applied to data
science-can unlock organizational value.
How the designers mindset can ground and
amplify analytic insights.

Introduction | v
A N A LY T I C S C U LT U R E

Why Hasnt Your Data Science

Investment Delivered on its
Promise?

The promise of game-changing results from data science has been touted
for years, and the allure of that promise has driven organizations across all
commercial markets and the public sectors to invest huge sums of money,
time, and resources to make it happen. Encouraged by the outcomes of
analytical pilots, organizations are now looking to scale data science
across their enterprise. As they continue their pursuit, they have found the
benefits are elusive and they are often humbled by organizational inertia.
The hard truth is that a key enabler to delivering on the biggest promise of
data science is transforming organizational culture. And, contrary to popular
belief, its not about transforming to just any culture. Organizations must
start to pursue an analytics-driven culture.

An analytics-driven culture uses analytics to generate
insights that can be used by organizations to inform
strategic decisions and propel the organization to the
next level of performance. One of the key benefits of
an analytics-driven culture is the difference in
timescales in findings solutions. With an
analytics-driven culture, the organization becomes
particularly adept at asking the right questions and
rapidly (within minutes sometimes) getting answers.
That shifts the time available to thoughtful discus-
sion on what analytical outputs mean and how they
inform decisions to derive desired outcomes.
 B O O Z A L L E N S D ATA S C I E N C E D E L I V E R Y M O D E L
Booz Allens Data Science Delivery Model describes + Generate Insights is where our experts apply
the ongoing activities that collectively evolve an data science and advanced analytics methods,
organization to an analytics-driven culture. We work tools, and techniques to support our clients in
with organizations to approach analytics in a holistic unveiling analytic insights.
matter through three phases: + Realize Value turns the analytic insights into
+ Imagine Outcomes focuses on deconstructing actionable results that enhance an organizations
business problems to define objectives that effectiveness, efficiency, and/or bottom line. We
will be met through analytics. We work with our support our clients in putting analytics into
clients to create analytics strategies, prioritize production, developing feed-back ecosystems,
initiatives, and establish a collaborative environ- establishing an environment of continuous
ment around analytics. learning, and building and managing their talent.

DATA SCIENCE DELIVERY MODEL

g
uildin ACC
sh ip B EL
ion ER
elat DOMAIN KNOWLEDGE AT
, R including industry expertise, familiarity with data sets, O
on RS
ti and operational understanding, underpins an
era organizations ability to maximize the value from its data
n

oll
Ge

ab
a

ora
Ide

tive

GENERATE
n,

IMAGINE
atio

INSIGHTS
nal

OUTCOMES
ritiz

ytic

What do I want to How do I unleash the

Prio

s,

achieve and how will I value from my data?

unity

get there? Apply data science to

ata
GE

TS

Identify business problems generate insights

ER H
ERATORS Opport

AT E I N S I G Develop, test and

aluation

and desired outcomes

Create analytic strategy evaluate hypotheses
and approach
, echnology Infusio

ANALYTICS-
DRIVEN
CULTURE
EL
ACC

n
IM

ES

G
A

IN M
E OUTCO
EA
UE
R

LIZ
E VA L

REALIZE VALUE
For my operations, organization,
bottom line, and culture
C Implement analytics in the enterprise
A

C g
EL Identify and apply insights to generate value
ER sin
en
AT al
OR ion
S ituat
ctiva
ted n rise,
alytics, daptive nterp

Analytics Culture | 7
 HOW TO INCUBATE AND SUSTAIN AN
ANALY TICS-DRIVEN CULTURE
It is no surprisebuilding a culture is hard and
there is just as much art to it as there is science.
It is about deliberately creating the conditions for
advanced analytics to flourish and then stepping
back to empower collective ownership of an organic
transformation. At Booz Allen Hamilton, we not only
have experience in helping our clients navigate this
challenge, but we also have experience from our
own transformation. Led in part by our 500+ data
science team, we began our own journey to an
analytics-driven culture, and, through our experiences,
we identified foundational building blocks that are
universal to anyone who wants to deliver on the
promise of analytics. While in the end each
organization will have its own unique flavor of its
final analytics-driven state, it is important to start
any journey by pursuing these foundational elements.
Weve outlined the basics for how to incubate and
sustain an analytics-driven culture.
R E C A L I B R AT E Y O U R T O L E R A N C E
F O R FA I L U R E FA I L U R E I S T H E
P R I C E O F D I S C O V E R Y:
To realize the full power of what data science can
provide, it is important to foster curiosity and
experimentation, and embrace failure for the
unexpected insights and learning that it can provide.
At Booz Allen, we created these conditions in a few
ways. First, we empowered our data science team to
investigate and discover; to design and run their own
experiments that blend inductive pattern recognition
with deductive hypothesis generation; and to explore
rabbit holes of interest. Recognizing, like many
others, that data scientists are rarely motivated by
money and prefer the time and space to geek out,
we established a series of reward and recognition
mechanisms focused on investing in passion projects.
Discovery has an added organizational benefit.

HOW TO INCUBATE AND SUSTAIN AN ANALYTICS-DRIVEN CULTURE

RECALIBRATE
YOUR TOLERANCE
FOR FAILURE
Failure is the Price of
Discovery
Y
CH

IN Scie ore ists

on y
N

Da bou a Sci
ati m

Tha
of Be RAR
HI TTE

CL nce
e

BE IV
ta t M ent
bo En

A at
nD

US
lla e
E
UR LA

Co th
r
YO F

E
an

is
It C

ANALYTICS-
DRIVEN
CULTURE
DE ANA e Neersant

E C NT A
M LY
Eve eel C

LU O
THAY I IDE
OC TI s to

B
F
r y on

ing
by llers
on ve

W OV
RA CS

Do
rn te
PR
TI

Lea tor y
ZE
d

MEASURE AND
REWARD
THE MEANS AS
WELL AS THE END
It Reinforces Behavior
that will
Deliver Results

8 | W H Y H A S N T YO U R D ATA S C I E N C E I N V E S T M E N T D E L I V E R E D O N I T S P R O M I S E ?
We found that discovery is self-fulfilling and that
once you start on the road to discovery, you gain
momentum and energy that permeates throughout
the organization, which provides powerful lift to your
data science team. Finally, it is important to note
that it is essential to fail fast in the discovery
process. It does no good to toil on something for
6 months and then learn that it doesnt work.
Experiment and fail quickly to discover more.
team and the broader data science community. At
Booz Allen, we have invested in tools, resources, and
platforms to empower data scientists to share their
work and build on each others successes. This
includes a sandbox to store algorithms and code so
that good work can be replicated without permission.
We also developed a tool that relies on collective
community participation to source and share data
sets that can be used for analysis.

B E I N C L U S I V E D ATA S C I E N C E I S A B O U T
M O R E T H A N D ATA S C I E N T I S T S :
What we learned in our journey is that true curiosity
and experimentation requires an enormous tent
of skills, abilities, and perspectives to achieve
meaningful outcomes for data science. It is
important to take an inclusive approach to advanced
analytics that extends beyond the traditional cadre
of world-class data scientists. For example, our
data science teams often include data scientists,
technologists, domain experts, organizational
design and strategy practitioners, design thinkers,
and human capital specialists. Diversity of thought
can help organizations unlock the particularly thorny
problems and extract value from their data.
F L AT T E N Y O U R H I E R A R C H Y I T C A N B E T H E
E N E M Y O F C O L L A B O R AT I O N :
Data scientists need the autonomy and the flexibility
to explore analytical challenges without the added
burden of unnecessary hierarchical structures. In
practice, this means providing data science teams
with access to resources, tools, and talent; the
freedom to scope and run their own experiments;
and the time and space to collaborate with a diverse
P R O V I D E A WAY I N T O T H E C L U B
STORY TELLERS LEARN BY DOING:
Data science and advanced analytics can be
intimidating for some and exciting for others.
Whats important is to provide a variety of learning
opportunities that can cater to the different skill
sets and attitudes that exist within the organization.
We tackled this challenge in a variety of ways:
(1) We sponsored analytical challenges and
hackathons both to test and satisfy the desire
of advanced practitioners while at the same time
providing a safe learning environment for those still
developing their skills; (2) We created a self-paced
online training course for data education, Explore
Data Science, to teach both foundational and
advanced data science concepts such as data
visualization; (3) Additionally, we created a data
science training program called TechTank to cultivate
skills through apprenticeship and structured training.
These are just a few examples. The bottom line is
that we learned it is important to design opportuni-
ties and learning programs at all levels to help
people get familiar with data science concepts and
become more comfortable in practicing them. Dont
cast data scientists as mythical rock stars and then

A K E Y P I T FA L L T O AV O I D
When building an analytics-driven culture, it is important to avoid creating high barriers to entry. Often organizations will
showcase top data science talent as an example of the capability. While it is intended to provide context and vision, it
can also be interpreted as an unrealistic comparison for some. Rather than illuminate a single data science rock star,
it is important to highlight a diversity of talent at all levels to help others self-identify with the capability. It is also a more
realistic version of the truth. Very rarely will you find magical unicorns that embody the full breadth of math and computer
science skills along with the requisite domain knowledge. More often, you will build diverse teams that when combined
provide you with the triple-threat (computer science, math/statistics, and domain expertise) model needed for the
toughest data science problems.

Analytics Culture | 9
One of the difficult aspects
of cultivating an analytics-
driven culture is that there
is a lot of noise on the topic.
10 | W H Y H A S N T YO U R D ATA S C I E N C E I N V E S T M E N T D E L I V E R E D O N I T S P R O M I S E ?
not offer a way into the club. Remember,
all data scientists have a role in providing
mentorship and learning opportunities.
Data scientists need to attend hackathons,
teach classes, and mentor junior staff
whenever possible.
D E M O C R AT I Z E A N A LY T I C S
EVERYONE NEEDS TO FEEL
C O N V E R S A N T:
For something to be part of organizational
culture, it must be part of the fabric of the
employee behavior. That means to incubate
and sustain an analytics-driven culture,
employees must interact with and use
analytics in their daily routines. In order for
this to happen, analytics must be available
and accessible to all employees in some
form or fashion. For example, at Booz Allen,
we developed a tool that allows users to
drag and drop packages of coding to help
solve analytical inquiries. The drag-and-drop
feature lowers the barrier to entry
and empowers everyday employees in
performing data analysis. Additionally, all
employees need a baseline of analytical
knowledge, starting with a common lexicon,
to facilitate productive collaboration and
instill confidence. While not everyone will be
data scientists, employees need to identify
with analytics and be equipped with the
knowledge, skills, and abilities to work with
analytics to drive smarter decisions and
deliver exponential organizational perfor-
mance. One key way we democratized
analytics at Booz Allen was through the
development of the Field Guide to Data
Science, which provides a baseline under-
standing about data science and a common
lexicon so that employees can communi-
cate with each other and engage in solving
analytical problems.
M E A S U R E A N D R E WA R D T H E
MEANS AS WELL AS THE END
I T R E I N F O R C E S B E H AV I O R T H AT W I L L
D E L I V E R R E S U LT S :
When building an analytics-driven culture,
it is important to reward the approach and
thought process that people take as well as
the results achieved. It is similar to when
partial credit was awarded in school
some credit was given if you took the right
approach even if you ended up at the wrong
answer. The same applies here. In order to
help shift mindsets and change behavior,
people need to believe in the analytical
process and thus be recognized for trying
it until it is customary.
At Booz Allen, we believe analytics
innovation starts at the staff level and have
implemented systems and processes to
reinforce that message. For example, when
we evaluate project success, we look at
more than just the expected results but
also the approach and learning outcome.
In one case, we were able to apply data
science to devise an approach that signifi-
cantly reduced the time it took to identify
fraud. While the results were impressive,
the innovative approach to the problem
became the star. In the end, we have found
that whether or not the results pan out, its
a win for useither weve discovered a
great new finding or we have inspired our
teams to continue looking.
DEBUNKING COMMON MY THS THAT MAY
PREVENT YOU FROM EVOLVING TO AN
ANALY TICS-DRIVEN CULTURE
One of the difficult aspects of cultivating
an analytics-driven culture is that there is
a lot of noise on the topic. In the above
section, we have isolated six common
cultural denominators that must exist for
an analytics-driven culture to thrive. In
reviewing various articles, press, and literature on
the subject, and comparing to our own experiences,
we also identified four common myths or misnomers
to help organizations avoid potential hazards. The
journey can be hard enough and its important to
focus resources on what counts.
MY TH #1: GOVERNANCE STRUCTURES
ARE THE SOLUTION:
Governance helpsorganizational bodies that
represent multiple stakeholder groups and meet
regularly certainly help facilitate collaboration and
information sharing. As any good data scientist
will tell you, however, correlation does not imply
causation. That is, simply because we find that most
successful analytics organizations have governance
structures in place, it is not the structure alone that
leads to the groups success. Participants in an
analytics function (e.g., data scientists, business
owners, operational experts) must work to establish
relationships and trust. This is what creates an
environment that is conducive to data transparency
and development of shared challenges, goals, and
priorities. Then, governance structures serve as
an incredibly valuable mechanism for ensuring
such topics are discussed with some regularity
and formality.
M Y T H # 2 : A N A LY T I C S O L U T I O N S
A N D O P E R AT I O N A L R E A L I T I E S A R E
M U T U A L LY E X C L U S I V E :
A common characterization we hear from clients is,
operational owners are stuck in their ways with no
vision, while data scientists dont understand the
complexity required to turn analytical answers into
reality. As with most entirely divergent perspectives,
the truth is found somewhere in the middle. Often,
we observe data scientists pushing operational
owners outside of their comfort zones when it comes
to exploring the realm of whats possible. The
responding hesitance is often the result of past,
failed experiences, or knowledge that implementa-
tion will require overcoming a series of operational
hurdles that can often take time. It is important that
the analytics community continues to push the
envelope, but also respect that change is a process,
and the benefit of time and careful planning can
prevent serious missteps and unintended conse-
quences. Prototyping is typically an effective way to
merge analytic solutions and operational realities.
Prototyping analytic solutions can generate organiza-
tional buy-in while helping to anticipate expected
implementation challenges to inform better planning.
M Y T H # 3 : D ATA S C I E N T I S T S P R O V I D E
A SERVICE TO THE BUSINESS:
In an analytics-driven culture, there is no customer
or provideronly partners. The term customer
implies that one party is a recipient of someone
elses service, which here, are analytic outcomes.
If this is the case, it is evidence that collaboration
is failing. Throughout all analytics endeavors, data
scientists and decision makers (or decision influ-
encers) should work with shared goals, equally
invested in both the process and the outcomes.
This means partnering every step of the wayin
developing hypotheses, determining appropriate/
acceptable data inputs, validating analytic
assumptions, and interpreting results. In this way,
we eliminate the us versus them mentality
and recognize that analytics is just part of doing
business, rather than an add-on or an input. An
added benefit is better business outcomes. As
stated earlier, inclusive teams are a primary
condition for any organization wanting to increase
organizational performance through analytics.
M Y T H # 4 : C O M P L E X A N A LY T I C S A N D
TOOLS ARE BETTER THAN SIMPLE
A N A LY T I C S A N D T O O L S :
We are excited about the technical power introduced
by tools like Hadoop and Spark, while techniques
like machine learning carry a certain sort of mystique
and excitement. Yes, modern tools and techniques
provide a world of opportunity that was once non-
existent, but lets not forget there is also beauty in
simplicity. The high-end, expensive tools are
powerful, but most organizations are only prepared
Analytics Culture | 11
 to use a fraction of their power. By over-purchasing
on analytical tools, an organization can unnecessarily
increase its analytical investment and make
it harder to prove a return on that investment.
Furthermore, the learning curve associated with the
high-end tools may make it harder for organizations
to realize the most value from their data. In the
end, information consumers must understand the
analytical insights from these tools with at least
enough certainty that they are confident in its output.
Embrace simplicity where simplicity will do, and save
complexity for the hairy problems. If the information
consumers understand the analytical results 95
percent of the time, perhaps they will just trust us
the other 5 percent.

WHEN DONE RIGHT, THE PAYOFF IS WORTH IT

Like most things in life, if it is hard, it is worth it.
Building an analytics-driven culture doesnt happen
overnight and it is not something that you can
force. As many can attest, organizational inertia is
powerful. However, by following these foundational
steps and taking care to avoid common myths and
pitfalls, organizations can begin a journey to set the
cultural conditions needed to deliver on the promise
of analytics. It is up to organizational leaders and
employees alike to build upon these elements and
design and grow their own version or interpretation
of an analytics-driven culture. Only then can the
promise of game-changing performance be
truly realized.

12 | W H Y H A S N T YO U R D ATA S C I E N C E I N V E S T M E N T D E L I V E R E D O N I T S P R O M I S E ?
 O R G A N I Z AT I O N A L P R O F I L E S

Getting Past the Bumps in the Road

With the explosion of data in virtually every aspect of society, a growing

number of organizations are seeking to take full advantage of analytics to
drive decision making. They are developing more sophisticated descriptive
and diagnostic analytics to understand what has happened in the past, and
why. They are poised to take analytics to an even higher level by using the
new wealth of data to predict what will happen in the future, and to
prescribe the best course of action.

No organization, however, instantly becomes adept
at this next generation of analytics. It is a journey,
and like any journey, there may be times when the
way ahead is not clear, and it becomes difficult to
move forward. Some organizations may be unsure of
the best path to take. Others may hesitate because
they are not convinced that continued investment in
analytics will pay off. Still others may find themselves
mired in data that is difficult to use.
Even organizations that are further along in the
journey can find their progress slowed. It may be that
they have achieved success in pilot and proof-of-
concept efforts, but have not figured out how to
translate that to a wider capability. Or, perhaps they
have all the elements in placepeople, processes,
and technologybut the larger organizational culture
still has not embraced the potential of analytics to
feed innovation.
These kinds of obstacles are natural and should be
expected. But they do not have to slow down an
organization for long. There are a number of ways
that organizations can restart their progress.
Booz Allen Hamiltons work with organizations
across the commercial and government sectors
found that organizations on the analytics journey
generally fit one of five profiles, each with its own set
of challenges. We have identified techniques that
can help all organizations move forward.
THE WILLING ORGANIZATIONCHARTING
A CLEAR PATH
This organization recognizes the potential of
analytics to create opportunities and solve complex
problems, but does not know exactly how to proceed.
Typically, the willing organization is early in the
journey, and may be using basic descriptive and
diagnostic analytics to investigate specific problems.
Now, however, it wants to broaden that capability
with analytics that can guide decision making, and
more fully support mission and business goals. But
what is the best approach? Without a clear direction,
any organizationno matter how willing and able
may find itself in a holding pattern.
Willing organizations often let their enthusiasm cloud
the need to fully map out all of the many elements of
an analytics capability. An analytics blueprint can
 help organizations understand how those elements
fit together, and what steps need to be taken in
what order.
For example, the blueprint can guide one of the most
effective steps: determining in advance where the
value in the data might lie. What is the art of the
possible? Instead of starting with a specific problem
or goal, start with the data itself. Look for trends and
issues that suggest how analytics can make a
significant difference in your organization. This
insight gives the organization a picture of what its full
analytics capability might look like, and sets the
context for creating the path to get there.
THE HESITANT ADOPTERUSING PROTOT YPES
TO OVERCOME DOUBT
This organization may be exploring more sophisti-
cated analytics, but is less sure of how to define its
value and so finds it difficult to move forward. While
the analytics visionaries want to increase invest-
ments in an analytics capability, they have not yet
gained the organizational support for doing so. The
organizations leadership, for example, may not be
convinced that analytics will really help the bottom
line. Or, individuals in the organization may not yet
recognize the value of analytics, and still prefer to
make decisions based on their own knowledge
and experiencedespite mounting evidence to
the contrary.
These types of concerns can be overwhelming,
stopping an organization in its tracks. One effective
solution is to think big, but start small. Organizations
sometimes believe that they must jump into the
next generation of analytics all at once. But if they
experiment with a limited number of prototypes,
they have a chance to see what analytics can do.
This approach does not require a major investment in
technology or people, and positive results from
prototypes will help build the business case for
new types of analytics and demonstrate its
potential value.
THE DATA-DISTRESSED ORGANIZATION
BRINGING THE STAKEHOLDERS TOGETHER
This organization is ready to develop more, increas-
ingly sophisticated analytics, but finds it difficult to
get its data all in one place in a form that is usable.
Perhaps the data is of poor quality, filled with errors
or incomplete. Or, there may be a reluctance to
share data across the organization. The data may
not be collected in a consistent way, or it might be

THE WILLING ORGANIZATION THE HESITANT ADOPTER THE DATA DISTRESSED ORGANIZATION

Characteristics Characteristics Characteristics

++ Beginning of the journey
++ Believes in the power
of analytics, but overwhelmed
with how to get started
++ Beginning/middle
of the journey
++ Analytics visionaries want to
invest more heavily in analytics,
but lack organizational support
++ Beginning/middle of the journey
++ Organization wants to develop
increasingly sophisticated
analytics, but is stymied by an
inability to get the underlying
data in order

What To Do What To Do What To Do

Chart a Clear Path: Explore data
to help inform the organizations
vision for analytics, and to chart a
step-by-step path to achieve that
vision
Use Prototypes to Overcome Doubt:
Think big but start small, using
prototypes to prove the value of
analytics to end users and help
overcome doubt
Bring the Stakeholders
Together: Collaborate and build
relationships between IT and
analysts to develop data that
can be used by both

14 | G E T T I N G PA S T T H E B U M P S I N T H E R O A D
locked in rigid data silos that are difficult to connect.
The list goes on.
Organizations can help get their data in order by
making sure that analysts are included in the data
management framework and design. Although an
organizations data usually belongs to IT, the analysts
are the usersand data problems can arise when
their point of view is not taken fully
into account.
A second method is to encourage collaboration and
relationship building through technical, organizational
and physical structures, such as new governance
policies, or activities in which teams from across the
organization work together to achieve a common
goal. Such structures help break down the barriers to
information sharing, which is the largest obstacle to
analytics at many organizations.
THE SCALING ORGANIZATIONESTABLISHING
A CENTER OF EXCELLENCE
This organization has achieved success using
sophisticated analytics to drive decisions in a
number of isolated efforts, but may have difficultly
effectively growing an organization-wide capability.
For instance, the organization may not yet have
figured out how to achieve economies of scale. Or, it
may have difficulty prioritizingif several stake-
holders are competing for limited resources, for
example, how does the organization choose which
analytics efforts will best advance the interests of
the mission or business? These and other challenges
often arise because organizations typically expand by
simply doing more of what they are already doing,
rather than leveraging their knowledge through
collaboration, and the sharing of best practices and
lessons learned. Establishing an analytics Center of
Excellence is one of the most effective methods of
achieving this task.
This Center of Excellence consists of a team that
sets standards and promotes collaboration
throughout the organization on a wide range of
topics, such as training. The team also collects and
disseminates best practices, such as analytical
approaches, tools and techniques. It also shares
lessons learnedtypically, knowledge that is
applicable from one effort to the next. With a Center
of Excellence, an organization can steadily grow its
analytics capability in an organic, efficient manner.

THE SCALING ORGANIZATION THE EARLY ADOPTER

Characteristics
++ Middle/far along in the journey
++ Analytics have led to value in
pockets of the organization,
but leaders are uncertain about
how to most effectively grow
capabilities
Characteristics
++ Far along in the journey
++ Analytics generate insights,
but not everyone has adopted
analytics to drive decisions,
nor do they look to analytics
to answer new questions

What To Do
Establish a Center of Excellence:
Stand up a team dedicated to
fostering collaboration, sharing
best practices, and setting
standards for analytics
What To Do
Initiate a Culture Change: Begin by
making the analytic models easier
for the end user to ingest and use

Organizational Profiles | 15
 THE EARLY ADOPTERINITIATING MOVING FORWARD ON THE JOURNEY
A CULTURE CHANGE Data, like any resource, exists only as potential
This organization is well established in delivering unless it can be tapped. Organizations that can best
predictive and prescriptive analytics, but it is still capture the value in their dataand use it in
trying to reach its ultimate goal: creating a culture of decision makingwill be the ones that thrive.
analytics-driven decision making. The organization Analytics is a rapidly expanding and dynamic field,
wants to use data and analytics to encourage and there is no single tried-and-true path to success.
curiosity among all its employees, not just in its Every organization will encounter bumps in the road
analysts. It hopes to inspire its entire workforce to as it finds its own way. But no matter where an
imagine how analytics might create exciting new organization is on the analytics journey, it should
opportunities for the enterprise, and solve its most think through where it is trying to go and why. What
difficult business and mission problems. does the next step look like? What does the end
A particularly effective method of achieving this state look like? By focusing on these kinds of
goal is to put data and analytics in the hands of questions, organizations can break free of their
end users from across the enterprise so they can current constraints and continue on their pathfrom
see the possibilities for themselves. Some advanced whatever point they may be along their journey.
technologies have visualization tools that enable this
kind of direct interaction and give end users the
flexibility to freely explore the data, following their
ideas and hunches wherever they may lead.
Such direct interactions will encourage end users
to think about how analytics might solve any
number of problems. This approach also increases
the transparency of the analytics, so that people
understand and trust them more.
It gives end users throughout the organization a
greater sense of ownership of the analytics and
inspires them to use data and analytics to help
drive their decisions.

16 | G E T T I N G PA S T T H E B U M P S I N T H E R O A D
 A N A LY T I C S & C H A N G E

Keys to Building Buy-In

Many organizations are poised to take full advantage of analytics to drive

mission and business successusing analytics not just to understand
past events, but to predict future trends and to prescribe optimal courses
of action. This ambitious goal requires more than the right technology,
people, and processes. Just as important is strong organizational buy-in
from everyone involved in the analytics capability, including both the owners
of the data and the analytics end users. This often requires overcoming
resistance on a number of fronts, from a reluctance to share information to
a hesitancy to use the analytics in driving business decisions.

Booz Allen Hamilton, a leading strategy and CREATING A VISION

technology consulting firm, helps clients create
To be successful, leaders must be able to clearly
critical buy-in through a change management
articulate why an analytics capability is important to
approach tailored to organizations that are standing
the organization and what the organization will look
up a capability to develop increasingly sophisticated
like when the capability is in place. Some members
analytics. Our experience with government and
of an organization may not be convinced that
commercial enterprises has shown that three
analytics will add value to the business, so it is up
aspects of change management are particularly
to leadership to make the case that it will be worth
critical. Vision is needed to help stakeholders
the effort.
understand leaderships long-term goals for using
analytics to drive decisions and actions. There must Even if you already started down the path of adopting
be stakeholder participation in the process to predictive and other sophisticated analytics, it is not
develop shared understanding of how analytics too late to define a vision, one that is:
can create value. And prototypes are needed to ++ Aspirational
demonstrate that the analytics vision can indeed
++ Based on an understanding of your organizations
become a reality. Together, these three elements
values
build buy-in for the analytics capability and bring your
organization together to work toward a common goal. ++ Reasonable to achieve
++ Mutually beneficial for everyone involved
 ANALYTICS CAPABILITY ADOPTION CURVE
OWNERSHIP
I am accountable for
participating in the
I am open-minded analytics capability
DEGREE OF SUPPORT

I understand how to the change and

the change tell others how it
contributes to the ADOPTION I confidently
contributes to the
analytics vision analytics vision share the successes
and how it related to the analytics
I am aware of capability stand up. I
the analytics impacts me ACCEPTANCE am committed to
capability sustaining the change
stand up and can clearly explain
UNDERSTANDING
how it relates to the
AWARENESS analytics vision

A good place to begin when defining the vision is a
stakeholder analysis. Whether you are planning to
stand up an enterprise-wide analytics capability or
provide analytics for a single business unit or
division, you should understand the cross-organiza-
tional concerns such as improved access to data
and data security.
A key to realizing the vision is communicating it to
stakeholders. This is most often achieved through a
strong leader at the top of an organization. But the
vision can also be realized from the ground up. For
example, Booz Allens visioning process helped an
intelligence agencyoverwhelmed by the volume of
data to be analyzedcreate a vision for their
analytics capability. That vision was communicated
primarily through well-respected domain experts at
the agency who led a grass-roots effort to build an
analytics capability and successfully influenced
leadership, staff, and peers. Whether the vision is
top-down or bottom-up, it is essential to identify
leaders and sponsors who can reinforce the ideas
and keep the effort moving. Booz Allen gives you, as
leaders and sponsors, the messaging and other
change management tools to define your vision for
analytics and present it to the organization.
Booz Allens collaborative visioning process helps
define a future state that reflects the unique realities
of your organization and cultivates a willingness to
participate. A key role of the vision is to help over-
come the roadblocks to buy-in. Among the ways to
accomplish this is by demonstrating leaderships
commitment and sponsorship, and by helping
analytics teams and end users establish shared
values and goals for the analytics. As these
efforts take shape through the vision, they must
be reinforced through participation in the analytics
capability and through prototypes.
PARTICIPATING IN THE ANALY TICS CAPABILIT Y
The next step to overcoming resistance and building
buy-in is to encourage stakeholder participation in
the analytics capability. One way to accomplish this
is to have stakeholders from across multiple busi-
ness units contribute to decisions on analytics tool
development. Through this engagement, stake-
holders become actively involved in making the vision
a reality. Participation calls for true collaboration
leaders do not simply push their ideas down, but
bring all the stakeholders into the effort. Through
regular feedback and other methods, stakeholders

18 | K E Y S T O B U I L D I N G B U Y - I N
can help shape and take ownership of the
organizations analytics capability.
The Analytics Capability Adoption Curve shows the
five basic stages of change adoption, from initial
awareness to ownership. Organizations can move
stakeholders along this path by establishing a
network of change agents at different levels and
domain areas. For example, change agents might
include mid-level managers from each business unit
or division, or representatives from analytics teams.
They are essentially embedded advocates that
facilitate communication among stakeholders.
Booz Allen helps develop this network of change
agents and gives them tools to target their
communications and feedback collection.
By fostering collaboration among the stakeholders,
the change agentsalong with leadershiphelp
break down the data silos and encourage information
sharing so that the results of the analytics will be
more complete and accurate. The stakeholders work
out their differences, offer suggestions, and under-
stand how they will all benefit. This same process
can help all those involved in the analytics capa-
bilityfrom the owners of the data to the end users
of the analyticsdevelop a set of shared goals. And
it can help make the data more transparent so that
users are more likely to trust the results.
GAINING BUY-IN THROUGH PROTOT YPES
As organizations develop greater analytics capabili-
ties to drive decision making, prototypes are
essential in solidifying the support that is built with
the vision and participation. Prototypes should focus
on areas where quick buy-in is needed most. For
example, if there is strong resistance to sharing
information, the prototype should be strategically
designed to show how the various stakeholders will
benefit. It is often helpful to start with smaller
projects or less complex analytics where quick wins
are more assured.
As illustrated in the chart below, prototypes that are
most likely to generate buy-in have four qualities:

TO BUILD BUY-IN, AN ANALYTICS PROTOTYPE SHOULD:

FEATURE A HARD PROBLEM If the problem is too easy to solve, people may not have confidence in the strength of the
TO SOLVE analytics capability. But make sure the problem is not insurmountable.

YIELD A HIGH RETURN Demonstration of business value is the holy grailthe analytics outputs must be
actionable, as people want to see results before they buy in.

Quick success, even if preliminary, is importantthe solution does not have to be perfect
PROVIDE AN ITERATIVE
to gain buy-in and inspire action. Be wary if the problem you are trying to solve requires
SOLUTION
extensive data collection or highly complex analytics to even begin to show results.

If the analytics capability is intended to serve multiple business units/divisions, the

REACH ACROSS TARGETED
problem you are trying to solve should resonate with all of them. Positive results for only
GROUPS
a small portion of stakeholders will limit organizational buy-in.

Analytics and Change | 19

OUR APPROACH
There are unique challenges to standing up a
sophisticated analytics capability, and a one-
size-fits-all change management approach will
not be effective. Booz Allen can partner with your
organization to implement a change solution
tailored to your specific needs and goals. Here is
the framework for our approach in the table below.
When standing up an analytics capability, organiza-
tional buy-in is just as important as the right data
and analytics. Booz Allen can help you create the
vision, participation, and prototypes needed
for success.
VISION, PARTICIPATION, AND PROTOT YPE:
A CASE STUDY
Vision: Booz Allens work with the Department of
Homeland Securitys Immigration and Customs
Enforcement (ICE) Enforcement and Removal
Operations (ERO) began with development of
analytics to support a new identification system.
As value was realized within the organization,
the engagement evolved to bring analytics to the
forefront of the organizations operations, which
established our clients as leaders in data-driven
decision making for all of ERO. The vision: to enable
ICE to align resources with demand, understand

CHANGE MANAGEMENT APPROACH FOR ESTABLISHMENT OF AN ANALYTICS CAPABILITY

PREPARING FOR CHANGE MANAGING CHANGE REINFORCING CHANGE

+ Understand the organizations + Develop, test, and evaluate + Continue operationalizing

analytics maturity and stakeholder
needs/concerns
+ Define and communicate the
analytics vision
+ Develop a change and
communication strategy
+ Create a multi-disciplinary team
(i.e., computer scientists,
mathematicians, and domain
experts) and cross-organization
change agent network
+ Identify and select prototype(s) with
the vision/end state in mind
analytical hypotheses and the analytics
integrate prototype lessons learned
+ Continue implementation of
+ Put algorithms into production; operating model and process
evaluate and refine changes
+ Determine tailored data + Establish mechanisms to enable
visualization methods supervised learning
+ Assess change impacts, update + Calibrate analytics success
change and communication strategy, instrumentation and
and provide training measurement
+ Establish incentive structure and + Establish pipeline of analytics
reward system, and implement priorities
operating model and process
changes
+ Expand the team/change agent
network
+ Establish mechanisms for
measuring and communicating
success

20 | K E Y S T O B U I L D I N G B U Y - I N
strategic and tactical gaps and potential
mitigation strategies, and improve mission
performance.
Participation: Booz Allen collaboratively
established an Analytical Hierarchy Process
(AHP), a system of voting to reveal the
organizations preferences, to help execu-
tive leaders engage in analytics tool
development and set overall direction for
the models. We also worked with ERO to
establish a Modeling Control Board (MCB)
to serve as the internal governing body for
the analytics tools developed in ERO. The
board consisted of stakeholders from
various units within ERO, which met
regularly to discuss model assumptions,
data issues (quality, availability, etc.),
model updates and changes, and resource
priorities. The board reviewed and validated
all significant components of model
development and usage and helped ERO
determine how to effectively invest in and
grow the analytics capability.
Prototype: Booz Allen assisted ICE ERO in
using quick-turn tools to develop proofs, not
masterpieces, in response to pressing
concerns by leadership. The iterative
approach provided a series of quick wins
with senior stakeholders and enabled the
organization to institutionalize analytics. At
the heart of this effort was a commitment
to data-driven decision making. Booz Allens
innovative spirit and strong partnership with
our clients advanced ERO from a strictly
reporting organization to an analytically
driven organization that proactively
manages its strategy, operations, budget,
and performance with data. Recently, the
Booz Allen team and client jointly received
the award for Innovation in Analytics from
the preeminent analytics organization in the
US, the Institute for Operations Research
and Management Science (INFORMS).
Booz Allen embedded
analytics at the core of
ERO decision making and
provided some of the best
and brightest individuals to
work on this ground-breaking
project. The team worked
closely with us to not only
develop innovative tools, but
also help us use these tools
in our strategic and tactical
decision making. As a result
of Booz Allens support, ICE
understands how its data
can help drive decisions
and results.
David Venturella, Former Director
of ERO and Director of Secure
Communities
Analytics and Change | 21
A L I G N I N G D ATA S C I E N C E

Making Organizational
Structure Work
As commercial and government entities develop data science capabilities,
an inevitable issue they face is how their data science teams should align
within the organization. Should the teams be centralized under a single
data science leader? Should they be dispersed throughout the organiza-
tion, permanently embedded in individual business units or divisions?
Should there be some combination of those two models, or perhaps yet
another alternative?

Organizations often adopt an approach without fully THE CENTRALIZED MODEL

considering the implications. But that can be risky
Centralized data science teams serve the entire
unless the data science capability is closely aligned
organization but report to a chief data scientist,
with the organizations size, diversity, culture, and
who decides which projects the teams will work
other factors, it may have a limited ability to drive
on, and how to manage the projects. Business
mission and business success.
units work with the data science teams to solve
While there are many possible organizational struc- specific challenges.
tures, we have found that three models typically work
This model often works best for organizations that
best for guiding where the data science capability
are operating with limited resources and have too
should reside and how it will support the business.
few data science experts to embed any in business
Each of the threethe Centralized Model, the
units for long-term assignments. With the centralized
Diffused Model, and the Deployed Modelhas
model, the chief data scientist can target projects
distinct advantages. Yet each also has potential
typically long-termthat offer the greatest benefit to
pitfalls that organizations must guard against. What
the larger organizations immediate needs. In
they all have in common is the need for data science
addition, because the data science teams are in
leaders and teams to be proactive in ensuring that
close contact with one another (often co-located),
the selected model works as intended. They must
they are likely to pool their knowledge so that
take the lead in fostering the necessary collaboration
lessons learned by one team can help others.
and communicationboth within the data science
teams and across the rest of the organization. One primary challenge in the centralized model is
that it can sometimes be difficult to establish trust
Here is a guide to the three primary data science
and collaboration between the business units and
alignment models and how to choose which one is
the data science teams. Business unit members
right for your organization.
may view data science teams as outsiders who
do not fully understand or feel invested in the
business problems, and they may see their relation-
ship with the data science team as us versus them.
To avoid this pitfall, data science leaders and teams
CHIEF DATA
need to take the initiative to create a collaborative SCIENTIST
environment. They must develop a partnership
mindset and demonstrate their commitment to BUSINESS UNIT
LEADS
helping the business units achieve their goals.
The chart below shows the advantages and
challenges of the centralized model, and lists
DATA SCIENCE
specific steps for making the model work. TEAMS

THE DIFFUSED MODEL Business units bring their problems to a centralized

Diffused, or decentralized, data science teams are
fully embedded in business units such as marketing,
research and development, operations, and logistics.
The teams report to individual business unit leaders
and perform work under their leadership.
This model often works best in organizations that
have large data science capabilities and the
data science team, overseen by a chief data scientist.
resources to embed teams in individual business
units for long periods on open-ended projects. A
benefit of this approach is that it allows data science
teams to gain a deepened understanding of how
analytics can benefit a particular domain or business

THE CENTRALIZED MODEL

ADVANTAGES CHALLENGES PLACES EXTRA FOCUS ON

+ Greater efficiency with limited
resources, including flexibility to
modify team composition during
the life of a project as needs
change
+ Access to data science is
organization-wide, rather than
limited to individual business
units
+ Central management stream-
lines business processes,
professional development, and
enabling tools, contributing to
economies of scale
+ Organizational separation
between the business units and
data science teams promotes
the perception that analytics are
objective
+ Project diversity motivates data
science teams and contributes
to strong retention
+ It can be difficult to enlist business
units that have not yet bought in to
data science
+ Business units often feel that they
compete for data science resources
and projects
+ Teams re-form for every new
problem, requiring time to
establish relationships, trust,
and collaboration
+ Business units must provide
another organization (i.e., the data
science unit) with access to their
data, which they are often reluctant
to do
+ As a separate unit with rotating
staff, data science teams may
not develop the intimate domain
knowledge that can provide
efficiency to future business
unit projects
+ Selling Analytics. Demonstrate
tangible impacts of analytics to
business unit leadersthey are
critical partners and need to buy in
+ Portfolio Management. Create
transparency into how the
organization will identify and
select data science projects,
including criteria to prioritize
opportunities and align resources
+ Teamwork. Establish early
partnerships between data science
teams and business units, which
will be integral to framing problems
and translating analytics into
business insights
+ Education. Train business unit
leaders on the fundamentals of
data science and the characteris-
tics of a good data science problem,
so people across the organization
can recognize opportunities

Aligning Data Science | 23

 unit. This is particularly valuable in diverse
organizations where certain mission areas or BUSINESS UNIT LEADS
product lines, for example, may require highly
specialized knowledge.
The primary challenge in the diffused model is that it
may be difficult for disparate data science teams to
share the knowledge they gain in the individual
business units. Data science needs to learn from
itself; it is imperative that insights, best practices,
and lessons learned are shared and incorporated DATA SCIENCE TEAMS
into the data science capability. However, when data
science teams are discrete, this can be hard to
Data science teams are fully embedded in business
accomplish, especially in organizations that are
units and report to individual business unit leaders.
already heavily siloed. To avoid this problem, data
science teams must establish mechanisms that will
make sharing an integral part of each teams The following chart shows the advantages and
workfor example, it might be established as a challenges of the diffused model, and lists specific
metric for success. steps for making the model work.

THE DIFFUSED MODEL

ADVANTAGES CHALLENGES PLACES EXTRA FOCUS ON

+ Data science teams can quickly
react to high-priority business
unit needs
+ Business units are more likely
to own the analyticsto be
involved with the data science
effort, accept the output, and
adopt some change as a result
+ Data science teams learn the
organizations data and its
context, reducing project
spin-up and helping them
become equal partners in both
solving problems and identifying
the possibilities
+ A deepened understanding of
the business inspires data
science teams to ask new, hard
questions of the data, and they
understand the right questions
to ask
+ Business units with the most money
often have full access to analytics
while others have nonethis may
not translate to the greatest
organizational impact
+ Data science teams may face
pressure to compromise their
objectivity to avoid making a
business unit look bad
+ Lack of central management may
result in redundant software
licenses and tools, which drives
up total costs to the organization
+ The structure offers limited
motivation for business units to
integrate, inhibiting collaboration
in already siloed organizations
+ Work may become stale to data
scientists, driving them to seek
new and diverse challenges
+ Governance. Establish cross-
functional group(s) responsible for
guiding organization-wide analytics
standards, to include data, tool
selection, and means of prioritizing
analytics efforts
+ Peer Collaboration. Establish forums
such as data science communities
of practice and mentorship circles
to share best practices and
lessons learned (e.g., trends,
algorithms, methods)
+ Creative Outlets. Fund analytics
competitions, crowdsourcing, and
conference attendance that allow
data scientists to exercise their
minds, solve new problems, and
explore techniques

24 | M A K I N G O R G A N I Z AT I O N A L S T R U C T U R E W O R K
THE DEPLOYED MODEL
As with the diffused model, data science teams are
embedded in the business units. The difference is
that the embedded teams in the deployed model CHIEF DATA
report to a single chief data scientist as opposed to SCIENTIST
business unit leaders. In this model, also called the
BUSINESS UNIT
matrixed approach, teams are generally assigned to
LEADS
individual business units, though they are sometimes
also assigned to broader product lines, or to mission
sets comprised of members from several business
units.
DATA SCIENCE
This model often works best in organizations with TEAMS
medium-sized data science capabilities ones
that have a sufficient number of teams to handle Data science teams are overseen by a chief data
multiple projects, but must still carefully target their scientist and forward deploy to business units.
resources. This model has many of the advantages
of both the centralized and the deployed models;
Because the deployed model is often seen as the
the data science capability is more of an organic
best of both worlds, organizations may be quick to
whole, yet the embedded teams are close to the
adopt this approach. But it is also the model with the
business units.

THE DEPLOYED MODEL

ADVANTAGES CHALLENGES PLACES EXTRA FOCUS ON

+ Shared benefits of both the
centralized and diffused model
+ Data science teams collectively
develop knowledge across
business units, with central
leadership as a bridging
mechanism for addressing
organization-wide issues
+ Access to data science is
organization-wide, and close
integration with business units
promotes analytics adoption
+ Project diversity both motivates
data science teams and
improves recruiting and
retention
+ Central leadership streamlines
career management approaches,
tool selection, and business
processes/approaches
+ Deployed teams are responsible to
two bossesstaff may become
uncertain about to whom they are
ultimately accountable
+ Data science teams may face
difficulty being accepted into
business units, where long-time
relationships have been established
+ Access to analytics-resources
may still feel competitive between
business units, and data science
units risk alienating business
units whose proposed projects
are not selected
+ Conflict Management. The chief
data scientist should proactively
engage business unit leaders to
prevent competing priorities from
becoming the data science teams
responsibility to resolve
+ Formal Performance Feedback.
Agree to performance goals at the
onset of each project, and collect
feedback during the life of project,
including at its conclusion
+ Rotation. Allow data science teams
to work on projects across different
business units, rather than within a
single business unittake
advantage of one of the main
benefits this model affords
+ Pipeline. Regularly communicate the
data science project pipeline,
allowing business units to see how
their priorities are positioned

Aligning Data Science | 25

 most risk. Deployed data science teams essentially
have two bosses, and conflicts inevitably arise. It is
not unusual for business unit leaders and data
science teams to disagree on how an analysis should
be conductedfor example, the priority in which
analytics efforts should be addressed. Typically, data
science teams get caught in the middle of doing what
is asked of them by the business unit leader and
what they feel is right by their own technical merits.
While they report to the chief data scientist, as a
practical matter they may be reluctant to go against
a business unit leaderparticularly one with whom
they work closely on a day-to-day basis.
The chart on the previous page shows the
advantages and challenges of the deployed model,
and lists specific steps for making the model work.
CONCLUSION
As organizations consider which model to choose,
they should also establish the mechanisms needed
to ensure communication and collaboration between
the various data science teams, and between data
science and business leaders. Regardless of the
model, the data science teams must be proactive
they are responsible for initiating and sustaining that
communication and collaboration.
26 | M A K I N G O R G A N I Z AT I O N A L S T R U C T U R E W O R K
Many factors go into choosing how to align data
science teams, including the size of an organization,
the diversity of its business or mission sets, its
culture and strategic goals, and its ability to hire and
retain data scientists. But it is important to note that
data science alignment is not a one-time activity.
As an organization and its data science teams
change and grow, the organizational structure may
need to change as well. Data science leaders should
periodically re-evaluate their current structure and
determine whether a different model would be
more effective.
Data science is an emerging field that offers exciting
new opportunities for commercial and government
entities. But success will not happen on its own.
Data science leaders and teams must chart
their own path and continually search for new
ways to improve.
 THE LE ADERSHIP ANGLE

Harnessing the Power of Data

Through The Stand-Up of a
Chief Data Officer

Across government and industry, enterprises are realizing that they may
be sitting on a pot of gold waiting to be discoveredtheir enterprise data
and the insights it can provide. Many of these enterprises are responding
by standing up Chief Data Officer (CDO) organizations to enable them to
maximize the value from their data. This is an emerging and evolving
role that varies depending on whether related roles already exist in the
organization (e.g., Chief Data Scientist or Chief Analytics Officer) and how
integrated data already is carrying out the organizations objectives
(e.g., whether data is the primary business/mission, data is a driver, or
data is a byproduct). Ultimately, a CDO must focus on the data assets,
understanding and marshaling them in support of the overall enterprise
strategy, and collaborating with other leaders (e.g., Chief Analytics Officer
and business line leaders) who will facilitate achieving the potential of
that data.

The best approach to fulfilling the CDO role is to be there are five driving principles that can help a CDO
two parts data evangelist and one part enforcer answer these defining questions outlined in the table
by setting policy and developing technology frame- on page 28.
works/guidelines while still allowing the data owners
and business lines to have the flexibility to imple- BE ROOTED IN THE ENTERPRISE STRATEGY
ment them in their own way. Starting small and
The CDOs objectives must be rooted in the value he
focused will enable the CDO to set the standard early
or she seeks to generate. Enterprise-level priorities
and evolve as the data and collaboration mature. In
and constraints (e.g., resources and regulations)
doing so, a CDO faces a unique set of decisions with
drive the need to identify the business and strategic
answers that are not black and white. Fortunately,
imperatives that are best met through the CDO

Align Data Use and Policies
With the Enterprise Strategy
A CDO must align data use and policies
with the overall enterprise strategy to
ensure that data sources are available
to support the business lines specific
questions, objectives, and hypotheses.
28 | H A R N E S S I N G T H E P OW E R O F D ATA T H R O U G H T H E S TA N D - U P O F A C H I E F D ATA O F F I C E R
versus other entities (e.g., business lines
and IT). A CDO should develop a strategy
that aligns with, and supports, the vision of
relevant enterprise strategies and a clear
intent for increasing the value data provides
to the enterprise. Developing the strategy
does not need to be a cumbersome or
immovable process. It can be as straight-
forward as defining an articulate goal
that serves as a guiding compass while
being flexible enough to allow for course
adjustments. That goal may focus on
increasing data access, preserving
equities, extracting more value from data,
or enabling the business.
SUPPORT AND AUGMENT THE EXISTING
ORGANIZATIONAL CONSTRUCT
The CDO must determine what functions to
perform and where in the organization he or
she will align. The CDO should keep in mind
only a few of the most important functions
should be focused on during the early
DEFINING QUESTIONS FOR A CDO
What value does the CDO bring, and how
visionary versus tactical should the
approach be?
What functions should the CDO perform, and
how should the team be aligned?
How does the CDO promote democratization of
data and analytics and balance the need to
ensure consistency and rigor?
How should the CDO engage with the enterprise
to drive toward a data-driven culture?
Once the CDO achieves quick wins and can
focus on long-term goals, how can the CDO
evolve the scope, role, and value?
start-up phase. Typical CDO functions can
include enterprise data governance, data
architecture, and data infrastructureeach
of which has many subfunctions (e.g., data
quality, metadata, document management,
content management, data warehousing,
reference data, and data security). These
functions should support the enterprise
architecture of the broader organization and
align to the needs of its core partners
including, business lines, IT, and data
owners. Determining these functions will
help delineate where the CDO should align
and how the different parts of the organiza-
tion will work together. Depending on the
organization, alignment to the Chief
Executive Officer may be necessary to have
the authority to serve as a data access and
governance enforcer or achieve the desired
cultural shift. At the same time, alignment
to the Chief Information Officer may
facilitate acquisition of the technical
infrastructure needed to achieve the CDOs
DRIVING PRINCIPLES
Be rooted in the enterprise strategy
Support and augment the existing
organizational construct
Ensure data access and governance
as an enforcer
Be a data evangelist
Scale it
goals. No matter the alignment, the
CDO must take the lead in fostering the
necessary collaboration and communication
to ensure alignment with the rest of
the organization.
ENSURE DATA ACCESS AND
GOVERNANCE AS AN ENFORCER
Well-defined governance processes and
consistent ways to share, access, and
manipulate data enable the enterprise to
develop reusable, intraoperative, and legally
compliant data capabilities in a short
amount of time. A CDOs focus should be
on setting a common definition of what
constitutes shared data and how it should
be moved across the enterprise. This
includes the initial and practical governance
or the basic rules of the road that will
provide the necessary guidance to allow
data to be tagged, shared, and accessed
and a data architecture and infrastructure
roadmap that is focused on enabling early,
quick wins. Ultimately, the business and
enterprise need to trust the data assets
so they can rely on the accuracy of the
derived analyses.
The policy set forth must also be imple-
mented through technology in a manner
that adheres to policy and is still
lightweight and flexible for data owners
and consumers. It should also take into
account best practices and lessons
learned from other partner organizations,
as appropriate.
BE A DATA EVANGELIST
The Trusted Data Format (TDF)
The CDO is a new and unfamiliar role for
most organizations. As the data evangelist,
Leveraging enterprise data headers,
it is the CDOs role to build a data-driven such as the Trusted Data Format (TDF),
culture across the enterprise. While some can reduce compliance and security
of this comes from the CDOs inherent role complexity by creating self-describing
as a facilitator of collaboration, equally or machine-learned data that prescribes
important are proactive means of helping how data systems must handle the
all layers of the organization, front-line multitude of organizational requirements
employees through executives, develop an placed on the data.
increased level of comfort with data
including collecting data, asking questions TDF remains flexible by providing the
option for the CDO to define the basic
of the data, and continuing to use data/
schema and policies that form the
analytics to drive decisions. To do this, a
enterprise standard while leaving the
CDO can identify and disseminate best
business lines and data owners to
practices and lessons learned while customize and provide further definition
understanding shared, cross-organization without being tied to a specific
areas of interest (e.g., most frequently used technology.
data, usage patterns, and techniques). A
CDO can also proactively communicate new
policies, guidance, and standards to all
relevant parties and sponsor committees
that work across the business lines to
respond to gaps in data management
processes and help resolve conflicts,
particularly around data usage and sharing.
In addition, a CDO can employ tools (e.g.,
drag and drop coding packages, reusable
visualization interfaces) that democratize
analytics by lowering the barriers to
entry for those not as skilled in
programming languages.
The Leadership Angle | 29

For CDOs, By CDOs
For CDOs, By CDOs: The Federal CDO
Council connects senior data and
analytics executives across government
in a trusted, collaborative setting to
share strategies on data and analytics.
The Federal CDO Council is member led,
following a grassroots agenda to amplify
the most promising opportunities and
address the most pressing challenges
faced by CDOs today.
30 | H A R N E S S I N G T H E P OW E R O F D ATA T H R O U G H T H E S TA N D - U P O F A C H I E F D ATA O F F I C E R
SCALE IT ++ Leveraging Innovation. To continue
evolving and promoting the enterprises
Starting small and focused allows the CDO
data-driven culture, a CDO can champion
to build the required foundation and buy-in
innovative activities such as internal
to scale and evolve his or her scope, role,
hackathons and idea incubators to
and value over time. Long-term initiatives a
evolve the way an organization is tackling
CDO may focus on include the following:
its most difficult data and analytical
++ Partnering to Promote Advanced challenges. These events
Analytics. In many cases, the people test and satisfy the desire of advanced
closest to the mission do not under- practitioners while at the same time
stand the possibilities of advanced providing a safe learning environment
analytics, whereas the people with for those still developing their skills.
advanced analytics skills do not
understand the mission well enough to A new CDO has the opportunity to set the
spot opportunity. In the evangelist role, standard. Starting small and focused on
the CDO is in an ideal position to partner these five driving principles will help a
with the business lines and analytics CDO make the defining decisions for the
teams to help them understand the organization and generate the momentum
art of the possible with advanced needed to garner support and further
analytics while ensuring analytics efforts grow a data-driven enterprise. The focus,
are connected to the so what or a role, and value of the CDO will then evolve
business problem and ultimate solution. over time to fully achieve the insights or
competitive advantage the organizations
++ Implementing Portfolio Management.
enterprise data can provide.
As the data-driven enterprise culture
continues to evolve and the organization
begins to pursue more applications of
data and analytics, portfolio manage-
ment becomes increasingly important
to identify existing capabilities, avoid
duplicative work, and prioritize future
endeavors. Portfolio management
includes governance and maintenance
of enterprise data capabilities and the
tools for users to leverage existing data
and analytics code for rapid develop-
ment of ad hoc analytics. Portfolio
management and the grassroots
innovation associated with disruptive
technology demand the CDO strike a
thoughtful balance between structured
process and the Wild West. Over-
structuring the management of data
and analytics capabilities will result in
lost innovation, whereas a lack of
structure can result in duplicative
efforts and lost focus.
 E V E R Y T H I N G YO U N E E D T O
K NOW ABOUT MANAGING
YO U R D ATA S C I E N C E TA L E N T
The Booz Allen Data Science Talent
Management Model
Recently, Harvard Business Review branded data
science the Sexiest Job in the 21st Century,
and the relatively new career field has been
growing in popularity since. Organizations are
clamoring to attract, hire, and build talent that
can unlock the power of their data and deliver on
its promising potential to increase organizational
performance and drive complex business deci-
sions. The hard truth, however, is that good data
scientists are hard to findcurrent demand far
outweighs supply. A recent survey by Gartner
found that over half of business leaders felt their
ability to carry out analytics was restricted by the
difficulty in finding the right talent. Furthermore,
not finding the right data science talent can have
a cost. It is not as simple as renaming a busi-
ness intelligence analyst, completing a day-long
training, or hiring any candidate with data
science in their resume. The National Business
Research Institute estimates the cost of one bad
hire can range from $25K to $300K depending
on job complexity.
THE RANGE OF TALENT REALITIES
Given the current demand for data
scientists, we often see organizations in
a few different states when examining
their data science talent reality:
+ The Aspiring Star: An organization that
is just beginning its data science
journey and needs to acquire and/or
hire talent
+ The Duckling: An organization with
potential data science talent that is
struggling to develop and transform
its staff
+ The Enviable but Vulnerable: An
organization that has data science
talent and must retain its much
sought-after resources.
To complicate matters, organizations
can transition organizational states at
different points across their analytical
journey, meaning that an organizations
talent challenges will also change over
time. To succeed in this type of dynamic
environment, a talent management model
can be a difference maker.
Booz Allen identified key activities, tasks,
and attributes required to successfully
perform in a data science role to help
clients understand the work to be
performed.
WHY THE JOB ANALYSIS MAT TERS:
+ Helps clients define and understand the
most important work of their data
science teams
+ Confirms the importance, frequency,
and relevance of tasks and knowledge,
skills, and abilities (KSAs)
+ Assists organizations to address
expansion of mission, establishment
of a new role, or an influx of new staff
+ Establishes the legal and scientific
basis
Booz Allen developed a scientifically valid
competency framework to help clients
determine what KSAs their data science
team needs.
WHY THE COMPETENCY FRAMEWORK
MAT TERS:
+ Provides clients with insights into what
characteristics and KSAs are required
for a data scientists individual success
+ Establishes a legally defensible
foundation for human capital practices
+ Useful when hiring/selecting talent,
developing the workforce, assessing
training needs, etc.
32 | T H E B O O Z A L L E N D ATA S C I E N C E TA L E N T M A N A G E M E N T M O D E L
Data scientists have a unique blend of talent requirements, Booz Allen took a
knowledge, skills, behavioral attributes, rigorous, scientifically based approach to
and personality traits that enable them to designing the foundation of a data science
combine and analyze massive amounts of Talent Management Model. To meet the
structured and unstructured data to solve needs of business leaders, human
some of an organizations most complex resources professionals, and our own data
problems. Their inquisitive nature, as well science workforce, Booz Allen started from
as their ability to implement the scientific scratch, leveraging the expertise of our
method and advanced tools and tech- human capital team and the knowledge,
niques, makes them exceptionally unique capabilities, and experience of our 500+
and vitally important in todays competitive data science team to describe core
environment. Yet few organizations have elements of data science work activities;
taken a proactive, rigorous approach to define those characteristics that best
defining the unique characteristics of data represent successful execution of the work;
science professionals, systematically and develop a Talent Management Model
documenting the type of work they perform; that will arm clients with the insights
the competencies required for success; and needed to best identify, attract, grow, and
the comprehensive approach to talent retain data science talent.
management required to identify, attract,
grow, and retain these professionals. DEFINING THE CORE ELEMENTS OF
Booz Allen Hamilton, a leading strategy and THE JOB: JOB ANALYSIS
technology consulting firm, accepted the First, Booz Allen began by conducting a
challenge. Recognizing the need to better comprehensive job analysis on our own
understand the key elements that make up data science cadre. With the largest
the data science role and data science
BOOZ ALLENS DATA SCIENCE COMPETENCY FRAMEWORK
Clusters Competencies
Technical: Knows How and Statistical Modeling; Research Design; Data Mining and
What to Do Integration; Data Visualization; Computer Science; Database
Science; Machine Learning; Programming and Scripting;
Information Assurance; Mathematics; Operations Research
Data Science Consulting: Domain Expertise; Business Acumen; Program Management;
Can Do in Client and Resource Allocation; Collaboration and Teamwork; Data Science
Customer Environment Consulting; Ethics and Integrity; Communications
Cognitive: Able to Do or Critical Thinking; Problem Solving; Inductive and Deductive
Learn to Do Reasoning
Personality: Willing or Perseverance; Innovation and Creativity; Resilience and Hardiness;
Motivated to Do Inquisitiveness; Adaptability/Flexibility; Ambiguity Tolerance; Detail
Orientation; Self-Confidence; Work Ethic
 number of data scientists supporting government
clients spanning every federal market sector in
addition to numerous commercial clients, Booz Allen
data scientists provide a rich and diverse cross-sec-
tion of the data science career field. The job analysis
identified both the criticality and frequency of specific
job tasks mapped across consulting, cognitive, and
personality areas, thereby solidifying their impor-
tance along with advanced technical skills. This kind
of analysis not only deepened Booz Allens under-
standing of the data science field but provided the
basis for defining core client talent needs through
developing a data science competency framework
and, later, the Booz Allen data science Talent
Management Model.
DEFINING CORE TALENT NEEDS:
COMPETENCY FRAMEWORK
Next, Booz Allen used the findings from our job
analysis to develop a data science competency
framework that encapsulates the KSAs for
successful job performance across data science
roles. To develop the competencies, we surveyed our
data science team and supplemented the findings,
via convergent validity, using a secondary job
analysis by Hogan Assessmentsan industry leader
in personality assessments. The findings indicated
there are four unique and complementary data
science competency clusters that, when considered
together, convey the essence of what it means to be
a successful data scientist.
The technical competency cluster depicts the
foundational technical and specialty knowledge and
skills needed for successful performance in each
job or role. The characteristics in the consulting
competency cluster can help data scientists easily
integrate into various market or domain contexts
and partner with business units to understand the
environment and solve complex problems. The
cognitive competency cluster represents the type
of critical thinking and reasoning abilities (both
inductive and deductive) a data scientist should
have to perform their job. Lastly, the personality
competency cluster describes the personality traits
that drive behaviors that are beneficial to data
scientists, such as inquisitiveness, creativity,
and perseverance.
Though there have been numerous technological
advances within this fast-growing field, to date, the
job analysis and competency framework are believed
to be the first, scientifically valid products to support
the design and development of a data science Talent
Management Model.
PUTTING IT TOGETHER: BOOZ ALLENS DATA
SCIENCE TALENT MANAGEMENT MODEL
Using the results of the job analysis and competency
framework, Booz Allen designed the remaining
components needed for a comprehensive data
science Talent Management Model. A Talent
Management Model is a suite of integrated tools that
help organizations strategically manage employees

Our deep human capital experience, coupled

HIGH QUALITY DATA with our unique understanding of the data
SCIENCE TEAMS science workforce can help clients win the
war for high quality data science talent.

TALENT MANAGEMENT MODEL

Who You Need: Where You Need Them: How You Keep Them:
Job Analysis Workforce Modeling Career Paths/Matrix
Competency Framework Workforce Design Training and Course Development
Position Descriptions Workforce Needs Retention Strategies
Hiring Criteria & Assessments Assessment Organization and Team Building

Booz Allen leveraged its human capital experts and its 500+ data
science team to conduct a comprehensive job analysis and develop a ORGANIZATIONAL MISSION
scientifically valid competency framework to help clients determine AND STRATEGIC OBJECTIVES
TALENT what knowledge, skills, and abilities their data science team needs.
AGEMENT
MODEL

Everything You Need to Know About Managing Your Data Science Talent | 33
 BOOZ ALLEN TALENT MANAGEMENT MODEL

Who You Need Job Analysis identifies key content in terms of activities and tasks
Key Benefits: involved and attributes required to successfully perform in a given job,
role, or position
++ Identify the data science talent that will
generate the most value
++ Rapidly define data science talent needs
Competency Framework encapsulates knowledge, skills and abilities
++ Select talent based on legally defensible for successful job performance
and scientifically valid data

Targeted Position Descriptions define and differentiate the primary

job responsibilities for a specific role within the organization

Hiring Criteria and Assessments identify candidates with the

essential ability, technical skills, and personality traits necessary for
successful job performance

Where You Need Them Workforce Modeling defines the type and amount of work within an
Key Benefits: organization with respect to the available talent supply and demand

++ Connect business goals to data science

talent requirements
Workforce Design describes the composition of the workforce and
++ Shape the data science workforce of
how work should be distributed across existing capabilities
the future
++ Mitigate talent risks before there is a
business impact Workforce Needs Assessment identifies and quantifies the need and/
or prevalence of desired skills or competencies to be obtained

How You Keep Them Career Paths/Matrix defines growth trajectories and transition points
Key Benefits: for personnel within the context of the organizations business
objectives
++ Reduce costly turnover in hard-to-find
data science positions
Training and Course Development creates developmental learning
++ Improve data scientists job satisfaction
opportunities to build new capabilities among junior talent and
and engagement
strengthen existing skill sets among senior talent
++ Motivate and grow data science talent

Retention Strategies provide organizations with the reward structure

and environmental factors to keep employees engaged and motivated

Organization and Team Building helps clients further develop

leadership skills and team dynamics to accelerate

34 | T H E B O O Z A L L E N D ATA S C I E N C E TA L E N T M A N A G E M E N T M O D E L
across the talent life cycle, including identifying, TALENT MANAGEMENT AS AN ENABLER
acquiring, developing, motivating, and retaining FOR SUCCESS
talent. The model is designed to be comprehensive
to provide the insight and tools needed to help Recruiting and selecting the right talent is a critical
organizations strategically manage talent and flexible component to building a robust data science work-
in allowing an organization to select and focus on force within an organization but it is only a piece
immediate talent needs, where appropriate. In short, of the talent strategy. To enable data scientists to
the model is designed to be customized depending flourish and deliver on their promise to enhance
on current organizational talent realities. organizational performance, organizations must
consider the full Talent Management Model,
The Booz Allen data science Talent Management including the amount and type of work to be
Model is not just a grab-and-go toolkit. Rather, its performed, the competencies and skills needed
a set of comprehensive service offerings that allow to perform the work, and the development and
clients to answer three key data science talent retention strategies the organization can employ
questions: Who do you need? Where do you need to support this kind of unique workforce. By imple-
them? How do you keep and develop them? When menting a comprehensive Talent Management
Booz Allen created the Talent Management Model, Model that addresses these critical components,
we intentionally developed service offerings to help organizations can achieve an engaged and
clients who are struggling to answer one or more successfully performing workforce, while gaining
of these key questions. Some of the foundational maximum returns on their analytical investment.
service offerings, such as the competency
framework and position descriptions, are designed
to be easily customized to an organizations needs.
Other offerings, such as workforce modeling, needs
assessments, and team building, entail greater
engagement and partnership with clients. Depending
on where an organization falls within the data
science talent spectrum, the Talent Management
Model will help inform which offerings are
most valuable for addressing its data science
talent challenges.
Implementing a Talent Management Model not
only helps data scientists understand their role
and career path within the organization; it helps
organizations establish, manage, and retain
their data science workforce in a strategic and
comprehensive manner.

Everything You Need to Know About Managing Your Data Science Talent | 35
T H E D ATA S C I E N C E C H A L L E N G E

How Design Thinking Can Help

You Realize Organizational Value

Organizations are operating in an increasingly complex environment. They

extend across traditional boundaries in both market reach and operational
footprint. Operations often involve messy and complicated human interac-
tions either through customers, partners, and/or employees. To further
complicate matters, organizations are tapping into increasingly sophisti-
cated technology capable of capturing oceans of data with varying degrees
of structure and quality. Many organizations have turned to data science
and advanced analytics to tackle todays complex organizational chal-
lenges. They still struggle, however, to capitalize on the power of analytics
to address their most important business challenges. Therein lies the data
science challenge. How does an organization focus the strength of its data
and analytics to create impactful organizational value?

Two key dimensions make it hard for organizations to
appropriately focus their data and analytics. First,
they need to have a deep and innate understanding
of their business. While this may seem obvious, the
reality is that this type of understanding is often
dispersed among organizational leaders, business
units, data scientists, individual employees, and
customers. It requires an insatiable curiosity and
institutional collaboration to discover. Without this
level of understanding, the organization will not be
able to identify and articulate the most pressing
business issues, and there is no way to ensure that
the organizations analytical horsepower is cohesively
focused on rich and relevant business questions.
Second, the organizational value desired from data
science is often found at the intersection of analytics
and context. While data science can provide predic-
tive and perspective analytics, insightful context can
explain the answer of why, including why does this
matter. Without the proper context, analytics will
only go so far.
CASE IN POINT DESIGN THINKING RISES TO THE DATA SCIENCE
Data scientists who are centralized in one govern- CHALLENGE: GROUNDING AND AMPLIFYING THE
ment agencys organization are empowered to ANALY TIC METHOD
prioritize their own projects. Often the data scientists
Solving the data science challenge is about an
use financial return as the primary criterion to
organizations ability to focus, embrace, and use
prioritize their efforts. This essentially means that
analytics to generate meaning and impact that can
projects focused on other strategic objectives, such
result in the next level of organizational performance.
as improving customer experiences or decreasing
One way to do this is to inject the art of design
propensity for operational errors, are deprioritized.
thinking into an organizations analytics approach.
Those responsible for such outcomes must often
Design thinking is a problem-solving and innovation
design and plan solutions without the benefit of the
methodologya tool box of techniques born from the
analytical insight that data science can provide.
designers mindset. It emphasizes solving problems
The risk here is that data scientists may expend
by starting with people (e.g., customers, employees,
their tremendous talent on questions that only
patients) rather than starting with technology or
serve pockets of the enterprise, rather than
business positioning.
delivering on the promise of data science that
can drive the collective enterprise to the next level In particular, design thinking can be a powerful
of performance. complement to data science, given its natural ability
to support the seamless shift between deductive
and inductive reasoning. Design thinking follows a

BOOZ ALLENS DESIGN THINKING METHODOLOGY

Immerse
Immerse Synthesize
Synthesize Ideate
Ideate Prototype
Prototype
Observe and
Observe and document
document Use discovered
Use discovered knowledge
knowledge Combine and
Combine and contrast
contrast Build, test,
Build, test and
anditerate
iterate
human experiences
human experiences to
to to reframe
to reframe the
the problem
problem and
and dissimilar information
dissimilar information to
to light and
light and lean
lean examples
examples
gain qualitative knowledge
gain qualitative knowledge shape understanding
shape understanding provoke unexpected ideas
provoke unexpected ideas

The Data Science Challenge | 37

Throughout Booz Allen similar method in terms of diverging and
converging in seamless succession to
Hamiltons methodology,
arrive at an increasingly finite focus. Design
there is an iterative flow of thinking pushes an organization to discover
activities that deliberately its most important business challenges,
helps to frame the right set of questions to
sequences divergent and
drive maximum value for the organization,
convergent steps. provokes exploration to expose critical
context to make sense of analytical
findings, enables collaboration to create
institutional knowledge, and drives testing
of resulting conclusions. By pairing these
like methodologies, organizations can not
only generate analytically sound solutions
but also provide a springboard for mean-
ingful impact, whether as new growth
opportunities or as areas for streamlining
processes for efficiency gains.
BOOZ ALLENS DESIGN THINKING TOOL BOX FOR ANALYTICS
BLEND DESIGN
RESEARCH INTO
ANALYTICS
Design research uses qualitative,
anthropologic techniques to uncover
human behaviors, unmet needs,
work-arounds, and context that
highly complements quantitative data
PROTOTYPE DESIGN
RESEARCH INTO
ANALYTICS
Creating fast and inexpensive prototypes
generates forward momentum and gathers
feed-back long before investments
are committed
38 | H OW D E S I G N T H I N K I N G C A N H E L P YO U R E A L I Z E O R G A N I Z AT I O N A L VA L U E
BOOZ ALLENS APPROACH TO DESIGN
THINKING KEY POINTS
++ Start from user needs, then factor in
business and technology requirements
++ Bring multiple perspectives to the table
++ Sequence discussions to fully diverge
before adding constraints
++ Make it realdont just talk about
solutions, draw them, build them, act
them out
Booz Allens design thinking methodology
takes a collaborative, system-level
perspective on a problem and incorporates
tools that lead to better problem framing
and reframing. Multiple perspectives and
collaboration among stakeholders creates a
more holistic understanding of the problem
HOST COLLABORATIVE
PROBLEM REFRAMING
WORKSHOPS
Reframing problems through tools and creative
collaboration among data scientists, business
owners, and other stakeholders ensures alignment
to meaningful questions
USE STRUCTURED
IDEATION TO
THINK BIG
Ideas dont come from thin air; they come
from prompts and inspirationusing
ideation exercises unlocks bolder thinking
that can then be toned down as needed
(i.e., context) and builds the institutional knowledge
and critical buy-in necessary to ensure ongoing
engagement and eventual scaling of capabilities and
solutions. These techniques refocus data efforts into
more meaningful and important questions that are
both business focused and analytically meaty.
Throughout Booz Allen Hamiltons methodology, there
is an iterative flow of activities that deliberately
sequences divergent and convergent steps. For
instance, instead of generating ideas in a linear
fashion, where each idea is offered and then
discounted for any number of reasons, in our
approach to design thinking, ideas are generated in
large batches, built on by others, and then prioritized
based on any relevant criteria. The result is a much
larger and more fertile sandbox of opportunity. These
solution development activities help data scientists
engage with business counterparts and work quickly
and creatively toward identifying and executing the
decisions and actions necessary to realize results
with the buy-in of key business partners.
HOW BOOZ ALLEN EMBEDS DESIGN THINKING
INTO ANALY TICS
Design thinking is both an end-to-end process and a
toolbox from which to pull tools and techniques for
modular application. As such, its integration with
data science can take several forms and requires
both experienced practitioners and sufficient training
of data scientists and other stakeholders to achieve
a shared mindset and language from which to
collaborate. At Booz Allen, because we believe it is
such a powerful complement, we train our own data
scientists and our clients in these techniques so that
they can get the most organizational value possible.
B L E N D D E S I G N R E S E A R C H I N T O A N A LY T I C S
One of the key aspects of design thinking is looking
for the hidden meaning or goals of the customer,
employee, partner, or patient, etc. Its not enough to
identify and understand a customers needorgani-
zations need to dig deeper. Establishing a design
research capability and conducting research in
sequence with quantitative methods of research
(e.g., surveys, multivariate testing, and digital
analytics) helps to generate a more complete picture
of not just whats happening, but why. This can
propel analytics organizations in new directions
through new levels of insight into problems that have
interactions among humans (customers, employees,
partners, etc.). The result can be a more fulfilling
analytical answer for all parties involved.
H O S T C O L L A B O R AT I V E P R O B L E M
REFRAMING WORKSHOPS
Booz Allens reframing workshops can bring together
data scientists, business owners, and even
customers (where appropriate) to explore and
discover the hidden roots of business challenges and
reframe problems into more meaningful
questions. Reframing workshops are designed to
challenge inherent assumptions made during the
analytical process, allowing the potential for break-
through thinking and solution development. Greater
value from data can be unleashed by following a
progressive cycle of analytical testing and reframing
to arrive at more promising (and elegant for that
matter) analytical solutions. Greater collective
understanding helps to design more insightful
research questions, and when paired with the right
analytical technique, increases the potential for
generating notable business impact.
U S E S T R U C T U R E D I D E AT I O N T O T H I N K B I G
Design thinking includes many techniques for
triggering ideas, drawing on existing patterns,
solutions, and concepts and reapplying them in
novel ways. With our design thinking techniques,
ideation moves from a critical linear process of
idea-constraint-idea-constraint to a sequenced
divergent process of generating a wealth of ideas
before converging on the most promising. These
techniques allow teams to turn insights from analysis
into so what actions necessary to move toward
organizational value.
The Data Science Challenge | 39
 R A P I D LY P R O T O T Y P E I N L O W - F I D E L I T Y WAY S
Design thinking generates low-fidelity prototypes to
demonstrate applicability and test ideas quickly and
cheaply before making significant investments.
These prototypes span mock-ups, illustrations,
paper-based interfaces, narratives, and other
techniques that are quick and low cost to produce
and easy to test with customers and stakeholders in
near-real time. As feedback is gathered, teams can
build smarter implementation roadmaps. For
example, one technique we practice is a cupcake
road mapping approach in which solutions are
planned in delightful portions with increasing
complexity added only when a solution proves
worthwhile. This allows organizations to dedicate
resources to only the most viable solutions that can
make a difference.
CASE IN POINT
A major supply chain company was struggling
with manufacturing data and determining the right
questions to ask of the data. The Booz Allen team
introduced design thinking as a technique to help
reframe the questions and hypothesis. Using design
thinking trained data scientists, the team conducted
a co-generative workshop to unpack the problem
and iterated until they determined a hypothesis
and variable to test. The workshop and subsequent
analysis were so successful that the client
launched two international centers to apply the
same techniques on a larger scale.

FINAL THOUGHTS
The data science challenge is not going away, and it
is often not as simple as creating an algorithm to
generate organizational value. Rather, Booz Allen
finds value by taking a holistic approach deriving
meaningful insights by combining data science with
innovative and creative methodologies such as
design thinking. Booz Allens approach for realizing
the real potential of analytics rests on this broader
view of capabilities that moves beyond algorithms to
creative and organizational approaches to institution-
alize data science as more than a stove-piped
function, but rather as part of organizational DNA.

40 | H OW D E S I G N T H I N K I N G C A N H E L P YO U R E A L I Z E O R G A N I Z AT I O N A L VA L U E
WERE HERE TO HEL P Angela Zutavern
zutavern_angela@bah.com
Booz Allen is one of the pioneers of data 703-984-1028
science. And now were the go-to firm for
Ezmeralda Khalil
business and government agencies developing khalil_ezmeralda@bah.com
a data science capability. Let us know how we 703-377-6269
can help you.

Josh Sullivan, PhD Angela Zutavern Peter Guerra

PA R TNER
@joshdsullivan
Leading our Data Science team
shows me every day the incredible
power of discovery and human
curiosity. Dont be afraid to blend
art and science to advance your
own view of data analytics. It can
be a powerful mixture.
VICE PRESIDENT VICE PRESIDENT
@angelazutavern @petrguerra
We may not realize it yet, but Data Science is the most fasci-
well all be data scientists in nating blend of art and math and
the future. code and sweat and tears. It can
take you to the highest heights and
the lowest depths in an instant, but
it is the only way we will be able to
understand and describe the why.

Ezmeralda Khalil Steven Mills Rick Whitford

PR INCIPA L PR INCIPA L PR INCIPA L
@ezmeraldakhalil @stevndmills @whitfordrick
The power of data science lies You really can change the world Save and analyze everything.
in the execution. through data science when you
have the right capability in place.
 Alex Cosmas Brian Keller, PhD Jamie Lopez, PhD
C H I E F D ATA S C I E N T I S T CHIEF TECHNOLOGIST S E N I O R A S S O C I AT E
@alexcosmas @brndnlkllr @jamielopezphd

Data scientists should be
truth-seekers, not fact-seekers.
Ill take passion over talent Fundamentally, people drive the
any day. analysis and make the interpreta-
tions, so it follows the more talented
your people then the better your
analysis and interpretations.

Patrick McCreesh Kristen More, PhD Ari Hamalian

S E N I O R A S S O C I AT E
@patrickmccreesh
Respect the data, bring on
the science.
S E N I O R A S S O C I AT E
@kristenmorephd
All the data and tools in the world
are meaningless unless you have
the right data science talent to
make sense of it.
L E A D A S S O C I AT E
@arihamalian
The cost of analysis will always be
cheaper than the price of ignorance.

Katie Hanley JD Hannick Susan Michener

L E A D A S S O C I AT E
@katiekara
To make data science sing, hit the
impact notes every day.
42
L E A D A S S O C I AT E
@JDHannick
Data doesnt solve problems,
people armed with data
solve problems.
L E A D A S S O C I AT E
@michenersusan
Change agents are force multipliers
cultivating them to help drive your
data science vision forward is time
well spent.
Erin Senter
L E A D A S S O C I AT E
@erinsenter
Building and managing data
science talent is the difference
maker in todays competitive
data market.
Kelly Smith Katie Wilks
L E A D A S S O C I AT E L E A D A S S O C I AT E
@kellemaries @boozallen
If I asked what my customers Theory is great but lets make
wanted, they would have said it real.
faster horses. - Henry Ford

Kat Wood Cutter Brenton Logan Gibson

L E A D A S S O C I AT E A S S O C I AT E A S S O C I AT E
@boozallen @boozallen @Rivannarvr
Its when the data influences an Dont drive business decisions Ignoring organizational culture
unprecedented decision, change, without a roadmap to a data will leave you data rich,
or impact that Im inspired most. science strategy. information poor.

A B O U T B O OZ A L L E N H A M I LT O N
Booz Allen brings its pioneering work in advanced analyticsand the industry-leading
expertise of its more than 500-member data science teamto transform our clients data
into actions that keep them competitive in todays data-driven economy. The first ever
National Data Science Bowl, along with Booz Allens recently launched Explore Data Science
training program and Field Guide to Data Science is part of the firms ongoing commitment
to supporting data science education and awareness. Booz Allen Hamilton celebrated its
100th anniversary in 2014 and continues to be a leading provider of management consulting,
technology, and engineering services to the US government and to major corporations.

43
www.boozallen.com/datascience